View metadata, citation and similar papers at core.ac.uk brought to you by CORE Available online at www.science.canterbury.ac.nz/nznsprovided by UC Research Repository Restoration of floodplain habitats for inanga (Galaxias maculatus) in the Kaituna River, North Island,

Peter M. Ellery & Brendan J. Hicks*

Centre for Biodiversity and Ecology Research, Department of Biological Sci- ences, School of Science and Engineering, The University of Waikato, Private Bag 3105, Hamilton 3240, New Zealand. *Corresponding author’s e-mail: [email protected]

(Received 31 July 2008, revised and accepted 23 March 2009)

Abstract

The Kaituna River on New Zealand’s east coast is about 50 km long and drains lakes and Rotoiti. An extensive flood control scheme, including stop-banks and gates, prevents flow of water from the lower river main channel to the original floodplain. In addition, a new, shortened channel (the Kaituna Cut) diverts the river directly to the sea instead of flowing through its original course into the Maketu Estuary. Minnow-trapping of inanga, common smelt, and bullies in 50 year-old ponds, recently constructed ponds, and riverine sites showed that new ponds were colonised by inanga within three months of construction. New ponds had higher catch rates than old ponds, and both new and old ponds had higher catch rates than the river, where no inanga were caught. Extrapolations from trap catch rates (2-13 inanga per 100 m2), suggest that old ponds held about 8-20 inanga per 100 m2, while in new ponds there were 24-67 inanga per 100 m2. Our study suggests that constructed ponds can provide suitable habitat for inanga from the Kaituna River to rear to adulthood, and shows the importance of off-river habitat for New Zealand’s most important whitebait species.

Keywords: inanga – Galaxias maculatus – habitat restoration – artificial ponds.

Introduction fish (Waters & Burridge 1999). Their diadromous juvenile stage (whitebait) Inanga, Galaxias maculatus (Jenyns 1842) enters New Zealand rivers principally is a globally widespread freshwater fish in spring (September to November). that is native to New Zealand, south- Whitebait runs in New Zealand have eastern Australia, South America and the declined in the last 40 years, and inanga Falkland Islands (McDowall 1990). The comprise most of the run in many rivers species has been characterised as one of (McDowall 1990). A combination of the world’s most widespread freshwater wetland drainage and flood control

New Zealand Natural Sciences (2009) 34: 39-48. © New Zealand Natural Sciences 40 New Zealand Natural Sciences 34 (2009) schemes have isolated inanga from their course into the Maketu Estuary directly lowland rearing habitats, and are likely to the sea via a new, shortened channel at to have played some part in their decline. Te Tumu known as the Kaituna Cut (Fig- The Kaituna River on the east coast ure 1). Little floodplain remains within of New Zealand is about 50 km long the stop banks of the lower Kaituna River, and drains lakes Rotorua and Rotoiti. It and various flood control measures have has been subjected to an extensive flood resulted in 67 km of stop-bank, 88 km control scheme that involves stop-banks of canals and drains, seven pump stations and gates to prevent flow of water from and five flood gate structures (Goodhue the main river channel to the original 2007). Since settlement in 1880, exten- floodplain. In addition, the entrance of sive wetlands such as the Kaituna Swamp, the river was diverted from its original the Waihi Swamp, and the Kaawa Swamp

Figure 1. Location of the Kaituna River and study ponds in the North Island, New Zealand. Source: Goodhue (2007). Ellery & Hicks: Restoration of floodplain habitats for inanga 41 have been reduced to 248 ha, or less than < 0.1 m s-1 (Jowett 2002). Overhanging 1 % of the original wetland (Ministry for vegetation and other forms of cover fur- the Environment 1997). The Kaituna ther enhance inanga habitat (Richardson River catchment covers 124,000 ha, with & Taylor 2002). Density dependent 48 % of this below the Lake Rotoiti mechanisms appear to be more important outlet at . After falling 260 m than recruitment for determining inanga through a steep narrow gorge, the lower abundance in their study stream (Jowett river watershed of the Te Puke Lowlands & Richardson 2003). includes the Mangorewa River, Waiari Constructed ponds have been a focus and Ohineangaanga Streams and Rapa- of recent inanga habitat restoration, e.g., rahoe and Kopuroa canals. Mean annual in the Te Wae Wae Lagoon area, South discharge of the Kaituna River at the sea Island (Smith 2004), and on the Waiau is 49 m3 s-1. River floodplain (Paterson & Goldsmith In 1990, spawning grounds for the 2002). Traps have been used to evaluate most common whitebait species, inanga the success of habitat restoration, and (Galaxias maculatus), were identified opinions have varied on the dependence on the lower Kaituna River. Generally, of catch rate on various types of bait (e.g., inanga are known to spawn within 1 Millar 2001, McDonald 2007). Few km of the upstream end of the saltwater evaluations of the effect of baiting have wedge (Mitchell 1990). In the Kaituna been carried out. River, inanga have been observed spawn- The objectives of this study were to ing on both banks about 2 km upstream evaluate the influence of bait on inanga from the sea in and around the location catch rates, to excavate new ponds next of the old ponds (the Borrow Pits) in to existing ponds on a small apron of 1988, 1989, and 1990 (Mitchell 1990). floodplain, to compare inanga catch Rearing inanga require low water veloci- rates in ponds and riverine sites, and to ties, avoiding velocities > 0.2 m s-1, but compare recruitment of inanga to newly have broad requirements for substrate constructed and older ponds as the mi- and water depth > 0.25 m (Lamouroux gration and rearing season progressed & Jowett 2005). Feeding velocities are (October to February).

Table 1. Water depths at high and low tides and approximate pond dimensions. Pond areas exclude the areas occupied by islands (30% of FNP, 25% of ENP, and 10% of LNP).

Site name Site type Code N traps Depth (m) Pond dimensions Width Length Area High tide Low tide (m) (m) (m2) River site River RS 10 1.5 - 2.2 0.4 Estuary pond Old pond EP 0 1.1 - 1.8 0 11 44 490 Small pond Old pond SP 5 1.3 - 2.0 0.2 11 29 250 Feeder channel Old pond FC 15 1.3 - 2.0 0.2 11 167 1830 First new pond New pond FNP 6 1.7 - 2.4 0.6 22 29 360 East new pond New pond ENP 6 1.6 - 2.3 0.5 27 36 560 Long new pond New pond LNP 8 1.6 - 2.3 0.5 22 84 1330 Total 50 42 New Zealand Natural Sciences 34 (2009)

Table 2. The influence of different baits on catch rates of inanga, common smelt, and bullies in artificial ponds in the lower Kaituna River floodplain between 29 June and 15 July 2007.

Bait N traps Inanga Common smelt Bullies Mean SE Mean SE Mean SE Cheese 16 7.56 3.30 1.31 0.68 0.38 0.20 None 16 2.81 1.26 0.38 0.20 0.63 0.24 Cat biscuits 16 1.56 0.77 0.06 0.06 0.63 0.24 Vegemite 16 1.25 0.59 0.13 0.09 0.88 0.34

Methods herb Polygonum sp., and reed sweetgrass Glyceria maxima. Study site Between 2003 and 2006, the Maketu Before excavation of the new ponds, Taiapure fenced off the floodplain sur- inanga spawning habitat was identified rounding the ponds from grazing ani- in the outflows of ponds known locally mals, cleared gorse (Ulex europaeus), and as the Borrow Pits, located on the true planted some native vegetation. Three right of the Kaituna River (NZ map new ponds (First new pond, FNP; East grid 2809675E, 6377740N, latitude new pond, ENP; and Long new pond, 37.75175oS, longitude 176.39910oE; LNP) were constructed between 12 and Figure 1). These ponds were created by 14 June 2007, and the culvert between the excavation of material to form the EP and FC was removed (Figure 1). This flood protection stop-bank in the mid isolated the area to be excavated from the 1950s. The area of the Borrow Pits is river, giving the lowest possible water level contained in a loop of riverbank on the for the period of the work and containing river side of the flood protection stop- sediment produced by excavation within bank. No spawning was observed at the the Borrow Pits area. The pond areas in Borrow Pits in the autumns of 2000 and Table 1 exclude the areas occupied by 2001, but in autumn 2002, pre- and islands (30% of FNP, 25% of ENP, and post-spawning fish were trapped (Young 10% of LNP). More detail on pond con- & Ellery 2002). struction with photographs is provided in The old ponds are relatively long and Ellery (2008). narrow (Figure 1). Estuary Pond (EP) Bait comparison trial and the Feeder Channel (FC) pond run parallel to the stop-bank, and water in A trial was set up in SP at the upriver FC is fed only through EP through a end of the Borrow Pits using 6 mm mesh culvert. The Small Pond (SP) has its own Gee minnow traps. SP was not affected connection to the river. EP completely by the excavations. Eight traps were set dewaters at low tide, but FC and SP per day for 30-60 minutes from 29 June retain 0.1-0.2 m of water depth at most to 15 July 2007 in the afternoon between low tides (Table 1). FC had dense growth 1230 and 1730 h. Because trapping of a mixture of submerged and emergent started in mid-winter, when inanga aquatic macrophytes, including the appeared to become less active, feeding oxygen weed Lagarosiphon major, parrot’s only briefly at dusk, most of this trapping feather Myriophyllum aquaticum, willow encompassed twilight. On each of eight Ellery & Hicks: Restoration of floodplain habitats for inanga 43

Table 3. Total catch in different habitats on five trapping occasions in the Kaituna River and floodplain ponds from 30 October 2007 to 13 February 2008.

Site type N traps Total catch (N fish) Inanga Common smelt Common bullies Eels New ponds 100 532 331 369 69 Old ponds 100 299 32 157 58 River 50 0 31 540 16 Total 250 831 394 1066 143 trapping occasions, two traps were set Evaluation of artificial habitat use with each bait type: cheese, cat biscuits, To evaluate the use of artificial habitat, yeast extract, or no bait as a control. Baits a combination of Gee minnow and were applied as a 2-3 mm thick slice of collapsible traps were set in old artificial a 40 by 80 mm block of Colby cheese ponds, newly dug artificial ponds, and crumbled into the trap, about 20 dry river sites within 100 m of the pond cat biscuits per trap, or a plastic sachet entrances. Traps were set overnight for 13- of Vegemite® yeast extract with the foil 17 h on 30 October, 30 November, and top removed. Over 8 trapping sessions, 16 December 2007, and 16 January and a total of 211 inanga, 29 common smelt 13 February 2008. During each trapping (Retropinna retropinna) and 40 bullies. session, 10 traps were set at river sites, 20 Both common bullies (Gobiomorphus traps were set in new ponds, and 20 traps cotidianus) and giant bullies (G. gobioides) were set in old ponds. In the river, traps were caught. Cheese as a bait caught more were set in pairs, each pair about 75 m than 3 times as many inanga as any other apart. Each pair was one Gee minnow bait type tested (Table 2), so all traps used trap and one collapsible trap set about in the habitat comparison were baited 4-5 m apart. Of the 20 traps in the old with cheese as described above.

Temperature (oC) 18 19 19 20 20 21 21 22 0.0 Kaituna River SP FNP 0.2

0.4

0.6 Depth(m) 0.8 Temperature Conductivity 1.0 0.4 0.5 0.6 0.7 0.8 0.9 1

Specific conductivity (mS cm-1)

Figure 2. Distribution of electrical conductivity and temperature with depth in two artificially created floodplain ponds and in the Kaituna River on 17 Feb 2008. N = 1 for each measurement. 44 New Zealand Natural Sciences 34 (2009)

Table 4. Mean lengths of inanga caught in old and new artificial ponds in the lower Kaituna River floodplain between 30 Oct 2007 and 13 Feb 2008. Analysis of variance (ANOVA) compares means for old ponds and new ponds for each sampling date.

Date Old pond New pond ANOVA N Mean SE N Mean SE F df P 30 Oct 07 2 62.0 1.0 6 85.7 7.6 2.878 1,6 0.141 1 Dec 07 40 76.3 3.3 17 73.2 4.7 0.258 1,55 0.614 16 Dec 07 37 78.1 2.8 71 68.3 2.0 8.450 1,106 0.004 16 Jan 08 21 78.2 3.7 86 76.3 1.9 0.204 1,105 0.653 13 Feb 08 43 90.8 2.7 41 76.1 2.7 14.905 1,82 <0.001 ponds, SP had five, and the long, narrow ponds compared to the adjacent river, FC had 15. In the new ponds, FNP had electrical conductivity and was tempera- six, ENP had six, and LNP had eight. ture profiles were measured once on 17 All traps were baited with Colby cheese. Feb 2008 in two ponds and the river with The circular entrance rings of all col- a YSI EC300 conductivity meter. lapsible traps were flattened to an oval of about 20 mm x 70 mm to minimise Results the entry of bigger common and giant Halocline bullies and eels (Anguilla australis and A. dieffenbachii). However, some predation Despite their shallowness (maximum of smaller fish in the traps was observed, depth 1.9 m), the pond FNP and SP so all fish in the traps, alive, dead, partial exhibited a marked haloclines, with but identifiable, were counted as trapped. warmer, saltier water on the surface than The trapping on 30 October 2007 was on the bottom (Figure 2). In contrast, a morning set (0700-1400 h, mean set the Kaituna River where the traps were time 7 h). After that, all trap sessions set (RS) was isothermal with a higher were overnight (set 1700-2050 h, hauled salinity close to the bottom than at the 0800-1300 h the next day, mean set time surface. This probably reflects a tidal 15 h). Traps were set on nights when wedge penetrating under the less dense there was a low tide evening and morn- freshwater. ing, and left to fish overnight, across the Evaluation of artificial habitat use high tide. This ensured that the traps were set in water at low tide and therefore The five trapping sessions in the artificially always submerged. All fish caught in traps created ponds caught a total of 831 were counted. Up to ten inanga from inanga, 394 smelt, 1,066 bullies and 143 each trap were measured to fork length. eels. No inanga were caught in the river. Before measuring, inanga were partially More fish were caught in new ponds than anaesthetised by immersion in melted ice in old ponds (Table 3). Low numbers of water. Recovery was almost instantaneous inanga were caught on 30 October 2007, when fish were returned to ambient water but numbers increased on 30 November temperatures. and thereafter (Figure 3A). In November, To investigate to extent of tidal flush- inanga catches were greater in pond ing and its influence on salinity in the habitats than in the river (Kruskal-Wallis Ellery & Hicks: Restoration of floodplain habitats for inanga 45

New pond Old pond River 16 A. Inanga

12

8

4 Mean catch rate (fish/trap) 0 25-Oct-2007 08-Nov-2007 22-Nov-2007 06-Dec-2007 20-Dec-2007 03-Jan-2008 17-Jan-2008 31-Jan-2008 14-Feb-2008 28-Feb-2008

New pond Old pond River 24 B. Bullies 20

16

12

8

4 Mean catch rate (fish/trap)

0 25-Oct-2007 08-Nov-2007 22-Nov-2007 06-Dec-2007 20-Dec-2007 03-Jan-2008 17-Jan-2008 31-Jan-2008 14-Feb-2008 28-Feb-2008

Figure 3. Mean catch rates (± SE) of inanga and a combination of common and giant bullies in old and new artificial ponds on the lower Kaituna River floodplain between 30 October 2007 and 13 February 2008.

New pond Old pond River 16 A. Smelt 12

8

4 Mean catch rate (fish/trap)

0 25-Oct-2007 08-Nov-2007 22-Nov-2007 06-Dec-2007 20-Dec-2007 03-Jan-2008 17-Jan-2008 31-Jan-2008 14-Feb-2008 28-Feb-2008

New pond Old pond River 3 B. Eels

2

1 Mean catch rate (fish/trap) 0 25-Oct-2007 08-Nov-2007 22-Nov-2007 06-Dec-2007 20-Dec-2007 03-Jan-2008 17-Jan-2008 31-Jan-2008 14-Feb-2008 28-Feb-2008

Figure 4. Mean catch rates (± SE) of common smelt and eels in old and new artificial ponds on the lower Kaituna River floodplain between 30 October 2007 and 13 February 2008. 46 New Zealand Natural Sciences 34 (2009)

P = 0.013). After October, bully catches into the first pond inlets that they find, were much greater in river than in the especially LNP, the most downstream ponds (Figure 3B; Kruskal-Wallis P < pond entry. The smaller mean size in new 0.001). From estimated pond areas (240- ponds late in the season (Feb 2008) was 1,410 m2) and the number of inanga probably caused by a school of inanga trapped, we can estimate the minimum entering the new ponds but not the old fish density. Mean catch rates of 2 fish/ ponds. trap in old ponds and 8 fish/trap in new ponds suggest that densities could not Discussion have been less than 2-4 inanga per 100 m2 in old ponds and 5-13 inanga per 100 Inanga in New Zealand commonly m2 in new ponds. Given the relatively few occupy lowland streams (Jowett 2002; traps that were put in each pond (5-15), it Richardson & Taylor 2002), but in the is highly likely that actual densities could lower Kaituna River flood protection have been greater. structures have restricted access of juvenile After initial large catches of common inanga to habitats inland of stop-banks. smelt in the new ponds in October, This has increased the importance of the catches were low in both pond types and limited floodplain habitats within the in the river for the remainder of the study stop-banks of the lower Kaituna River for (Figure 4A). Catches of eels (primarily rearing and spawning inanga. In view of shortfin; Anguilla australis) were always the rapid occupancy of newly constructed low (Figure 4B), probably because the ponds by inanga (within 3 months after small size of the trap entrances restricted construction), our results suggest that entry. inanga in the Kaituna River will readily use artificially created ponds, and their Length analysis continued occupancy in the ponds shows From December 2007 to February 2008, that they can provide suitable habitat for inanga were larger in the old ponds the rearing to adulthood. than in the new ponds (Figure 5). These We can extrapolate from trap catch differences were significant only on 16 rates on the basis that Gee minnow traps December 2007 and 13 February 2008 have been found to catch ~ 20 % of fish (Table 4). Differences in sizes were most present (McDonald 2007). Using this likely caused by inanga moving upstream 20% figure, we estimate that in old

100

90

80

70 Old pond Meanlength (mm) 60 New pond 50 25-Oct-07 14-Nov-07 4-Dec-07 24-Dec-07 13-Jan-08 2-Feb-08 22-Feb-08

Figure 5. Mean lengths (± SE) of inanga caught in old and new artificial ponds on the lower Kaituna River floodplain between 30 October 2007 and 13 February 2008. Ellery & Hicks: Restoration of floodplain habitats for inanga 47 ponds, there were ~ 8-20 inanga per 100 Enhancement subcommittee of the m2, while in new ponds there were 24-67 Maketu Taiapure Trust and with the inanga per 100 m2. This exceeds the den- assistance of Kim Young from New sities estimated in ponds in the Te Wae Zealand Department of Conservation. Wae inanga restoration, in which spot- Advice was given by Ray Bushell and lighting revealed up to 0.28 inanga per Charles Mitchell. Funding was obtained 100 m2 (Paterson & Goldsmith 2002). from The Rainbow Warrior Fund, Te In Paterson & Goldsmith’s study, inanga Kotahitanga O Te Arawa Waka and were found in riverine sites at densities Environment ’s Environment of 0.0002-0.807 inanga per 100 m2. In Enhancement Fund. An extra day of our study, no inanga were caught in the digger time was kindly donated by Reads Kaituna River, suggesting that riverine Transport Ltd of Paengaroa, and we thank habitat in the main river was less suitable Dion Henderson for his careful digger than constructed ponds, which further operation. The excavation work is covered shows the importance of off-river habitat by Resource Consent No. 63887, granted in the Kaituna River for New Zealand’s in March 2007 by Environment Bay of most important whitebait species. Low Plenty. We thank two anonymous referees water velocities are preferred by inanga for their helpful comments. (Lamaroux & Jowett 2005), and this might be the limiting factor in the Kai- References tuna River. That the minnow traps in the river caught bullies and smelt successfully Ellery, P. (2008). Restoration of inanga shows that the absence of inanga is a reli- rearing habitat in the Kaituna River able result. margin. Unpublished Graduate Pond construction on lowland flood- Diploma in Applied Science plains is a viable way of restoring inanga dissertation, University of Waikato, populations. Salinity and temperature Hamilton, New Zealand. profiles show that brackish water is Goodhue, N.D. (2007). Hydrodynamic flushed into the ponds by the tides, while and water quality modelling of the a layer of water with lower temperature Lower Kaituna River and Maketu and salinity remains at the bottom of the Estuary. Unpublished MSc thesis. ponds. Our project shows that inanga, University of Waikato, Hamilton, and associated species such as smelt and New Zealand. eels, will readily occupy constructed Jowett, I.G. (2002). In-stream habitat pond habitat that is connected to the suitability criteria for feeding inanga river, even newly constructed ponds. The (Galaxias maculatus). New Zealand question that remains how much habitat Journal of Marine and Freshwater restoration will be required to enhance Research 36: 399–407. the whitebait run significantly. Jowett, I.G. & Richardson, J. (2003). Density-dependence and population Acknowledgements variability of inanga (Galaxias maculatus (Jenyns)) in a small New This work was undertaken as part of a Zealand pastoral stream. Australian Graduate Diploma (Biological Sciences) Society of Fishery Biologists annual at the University of Waikato. The project conference, 29 June-4 July 2003, was initiated by the Monitoring and Victoria University, Wellington, New 48 New Zealand Natural Sciences 34 (2009)

Zealand. Invercargill, New Zealand. Lamouroux, N. & Jowett, I.G. (2005). Smith, C. (2004). Survey of Whitebait Generalized instream habitat models. Habitat Enhancement Ponds at Te Canadian Journal of Fisheries and Wae Wae Lagoon. Unpublished Aquatic Sciences 62: 7–14. report for Diploma Environmental McDonald, A.E. (2007). Improving Management. Southern Institute the success of a translocation of of Technology, Invercargill, New black mudfish (Neochanna diversus). Zealand. Unpublished MSc thesis, University Richardson, J. & Taylor, M.J. (2002). of Waikato, Hamilton, New Zealand. A guide to restoring inanga habitat. McDowall, R.M. (1987). The occurrence NIWA Science and Technology Series and distribution of diadromy among No. 50. National Institute of Water and fishes. American Fisheries Society Atmospheric Research, Wellington, Symposium 1:1-13 1987. New Zealand. McDowall, R. M. (1990). New Zealand The Ministry for the Environment. freshwater fish: a guide and natural (1997). The state of New Zealand’s history. Auckland, Heinemann Reed. environment 1997. The Ministry for Millar, J.P. (2001). Inanga in the the Environment. Wellington, New lower Waikato River. Unpublished Zealand. MSc thesis, University of Waikato, Waters, J. M. & Burridge, C.P. (1999). Hamilton, New Zealand.Unpublished Extreme intraspecific mitochondrial MSc thesis, University of Waikato, DNA sequence divergence in Galaxias Hamilton, New Zealand. maculatus (Osteichthyes: Galaxiidae), Mitchell, C.P. (1990). Whitebait one of the world’s most widespread spawning grounds in the Bay of freshwater fish.Molecular Phylogenetics Plenty. New Zealand Freshwater and Evolution 11: 1-12. Fisheries Miscellaneous Report No. Young, K. & Ellery, P.M. (2002). 40, Freshwater Fisheries Centre, MAF Whitebait spawning in the Kaituna Fisheries, Rotorua, New Zealand. River borrow pits. Preliminary Paterson, R. & Goldsmith, R. (2002). report. Bay of Plenty Freshwater Fish Te Wae Wae whitebait habitat survey Report (02/03). Unpublished report, March 2002 and recommendations Department of Conservation, Rotorua 2001-2002. Report prepared for New Zealand. Waiau Fisheries and Wildlife Habitat Enhancement Trust. Unpublished report, Waiau Fisheries and Wildlife Habitat Enhancement Trust,